3D semantic segmentation and classification for assisted indoor navigation

This thesis presents an approach to indoor autonomous navigation, using techniques like 3D reconstruction, semantic segmentation, and Simultaneous Localization and Mapping (SLAM). By capturing and understanding indoor environments, the system aims to provide assistive navigation for individuals with disabilities (state-of-the-art ). The proposed method utilizes RGB-D sensors to generate dense point clouds. SLAM algorithms are used to create maps of the environment, enabling precise localization and path planning while machine learning algorithms are applied to classify these segmented spaces by creating an user-friendly map.